Update Dask Array's FFT implementation to include norm keyword.

[theXYZT]

Numpy/Scipy FFT functions now accept a norm keyword for normalizing the output. Currently, dask.array's FFT implementation does not allow for this. It would be great to update Dask Array's implementation to match with the current Numpy/Scipy API.

Here's a snippet of what currently happens:

In [1]: import numpy as np

In [2]: import dask.array as da

In [3]: np.fft.fft(np.ones(8), norm='ortho')
Out[3]:
array([2.82842712+0.j, 0.        +0.j, 0.        +0.j, 0.        +0.j,
       0.        +0.j, 0.        +0.j, 0.        +0.j, 0.        +0.j])

In [4]: np.fft.fft(da.ones(8), norm='ortho')
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-4-f57d26876801> in <module>
----> 1 np.fft.fft(da.ones(8), norm='ortho')

<__array_function__ internals> in fft(*args, **kwargs)

~\.conda\envs\main\lib\site-packages\dask\array\core.py in __array_function__(self, func, types, args, kwargs)
   1530         if da_func is func:
   1531             return handle_nonmatching_names(func, args, kwargs)
-> 1532         return da_func(*args, **kwargs)
   1533
   1534     @property

TypeError: func() got an unexpected keyword argument 'norm'

[theXYZT]

This was requested in #3280. I didn't see it in my search, my apologies. I'll close this issue now.